Intel’s Factory Expansion Advances as AI Chip Demand Tests Its Foundry Ambitions

Table of Contents

• You might want to know


• Main Topic


• Key Insights Table


• Afterwards...

You might want to know

• What do the latest construction milestones reveal about Intel’s plans to expand its U.S. chipmaking capacity?


• Could strong AI-related demand and tight advanced manufacturing capacity at Taiwan Semiconductor Manufacturing Company (TSMC) help Intel attract major outside customers?

Main Topic

Intel (INTC-US) has reached new milestones in its U.S. factory expansion. Equipment installation has begun at Fab 62 in Arizona, while the company’s new Ohio site has completed installation of its first large roof truss. The developments suggest that a long-delayed expansion program is gradually moving ahead as demand for chips rises alongside artificial intelligence (AI). However, construction progress alone will not determine whether these projects generate an adequate return. Intel still needs to attract enough external foundry customers to use the capacity.

The distinction between building factories and filling them is central to the company’s plans. A modern fabrication plant requires substantial investment, and equipment installation is a major step toward production as well as one of the most expensive phases. Intel’s Arizona progress is comparatively advanced, while the Ohio project remains at an earlier stage. At both sites, the commercial question is whether future demand—particularly demand from customers outside Intel—will be strong and dependable enough to support the cost of expansion.

Arizona equipment installation moves forward

Naga Chandrasekaran, Intel’s foundry chief, said on Wednesday (21st) that equipment installation is underway at Fab 62, near Phoenix. The facility is expected to begin operating in 2028. This milestone indicates that the project has moved beyond construction alone and into the process of preparing the factory for manufacturing.

Fab 62 is located near Fab 52, which is already operating. Fab 52 began mass production of chips using Intel’s advanced 18A process at the end of 2025. Those chips are currently used mainly in Intel’s own processors. The existing facility provides a reference point for the Arizona campus, but it does not by itself demonstrate that Intel has secured large outside customers for the 18A process.

Installing equipment is an essential part of preparing a semiconductor factory to produce chips. It is also an exceptionally costly undertaking: an individual piece of equipment can cost hundreds of millions of dollars. Consequently, each move toward installation represents both operational progress and a major commitment of capital. The timing and utilization of the equipment will matter to the eventual economics of the site.

Ohio construction remains behind its original schedule

At Intel’s New Albany, Ohio, site, work is still focused on the building structure. The installation of the first large roof truss marks a new stage in the construction process, but the project remains much less advanced than equipment installation in Arizona.

When Intel announced the Ohio project in 2022, it originally planned to begin delivering chips in 2025. The target for completing the first factory has since shifted to 2030, while the second is expected to be completed in 2031. The revised schedule places the first facility at least five years behind the original plan.

The extended timeline matters because a factory’s expected output and customer commitments are tied to when it can begin production. A delay can push back revenue and make it harder to assess when an investment will pay off. It also gives the company more time to secure demand, although that time does not guarantee that customers will ultimately commit to production orders.

In September, reports suggested that SK Hynix (SKHY-US) might lease part of the Ohio facility to produce memory. Neither company confirmed the reports. The possibility therefore remains unverified and should not be treated as a secured order or a confirmed customer agreement. For Intel, confirmed commitments would be a more meaningful indicator of commercial progress than speculation about potential use of the site.

AI demand supports Intel’s near-term capacity needs

Intel CEO Lip-Bu Tan took office in March 2025 and adopted a more cautious approach to the large-scale expansion strategy pursued by his predecessor, Pat Gelsinger. Tan emphasized that new factories should move forward only when the company has confirmed sufficient demand. The recent construction and equipment milestones suggest greater confidence in future capacity needs, but they do not erase the importance of disciplined investment decisions.

The clearest demand momentum currently comes from Intel’s own products. Orders for data-center processors used in AI services have surged, and the company is working to catch up with demand. This provides a direct reason to expand manufacturing capacity, even before large external foundry orders are secured. It also illustrates that Intel’s factory plans serve more than one purpose: they can support the company’s internal product requirements as well as its ambition to manufacture chips for other businesses.

In July, Intel raised its budget for factory and equipment spending this year to approximately $20 billion and said it expects spending to increase significantly in 2027. The planned investment remains below Wall Street estimates that TSMC will spend approximately $60 billion. The comparison highlights the scale of capital required to compete in advanced semiconductor manufacturing, while also showing that the companies are making investment decisions at different levels.

AI-related demand can help justify additional capacity, but it does not automatically solve Intel’s customer-acquisition challenge. Demand for Intel’s own processors may support factory utilization, yet a successful foundry business also depends on external customers trusting Intel’s processes, delivery schedules, and manufacturing capability.

External foundry orders remain the key test

Intel has not yet secured a major external customer for the 18A process in Arizona. Customers considering a foundry partner typically need confidence that a process can meet their technical requirements and that production can be delivered reliably. Intel’s previous delays involving its 7-nanometer and 10-nanometer processes have contributed to caution among potential customers, who may be reluctant to make another substantial commitment without evidence of consistent execution.

Nonetheless, tight advanced manufacturing capacity at TSMC could create an opening for Intel. Mizuho Securities analyst Jordan Klein has pointed to intense competition among major companies seeking capacity at TSMC, including Nvidia (NVDA-US), Advanced Micro Devices (AMD-US), Apple (AAPL-US), and Broadcom (AVGO-US). When several large chip designers compete for limited manufacturing slots, alternative sources of production can become more attractive.

TechInsights analyst Dan Hutcheson has said that customers want a second source of supply, which can also strengthen their negotiating position with TSMC. A second source can provide greater supply flexibility and reduce dependence on a single manufacturer. It may also give customers more leverage in discussions about capacity and commercial terms.

Hutcheson said that several major companies are already conducting tests at Intel. These evaluations are an encouraging sign of interest, but testing is not the same as a production contract. A customer may assess a process and still decide not to proceed, or it may take time to move from technical evaluation to volume manufacturing. The decisive milestone will be whether testing turns into confirmed, repeatable production orders.

For that reason, Intel’s expansion should be assessed through two separate measures. The first is physical progress: construction, equipment installation, and readiness to manufacture. The second is commercial progress: customer qualification, production commitments, and sustained use of the factories. The Arizona project has reached an important equipment milestone, while Ohio is advancing through structural construction. Whether those facilities become financially successful will depend on how effectively Intel matches capacity with real demand.

Key Insights Table

Afterwards...

Intel’s factory milestones point to a broader challenge facing the semiconductor industry: building capacity is only one part of creating a resilient and competitive supply chain. As AI workloads increase demand for computing, companies will need to understand how much manufacturing capacity is required, where it should be located, and how to avoid investing ahead of durable customer demand.

Further exploration of advanced process technologies, manufacturing automation, equipment productivity, and supply-chain resilience could help clarify how new factories can reach stable production more efficiently. The industry will also need transparent ways to assess whether a facility can meet customers’ performance, quality, and delivery needs. These capabilities matter not only to Intel, but also to chip designers seeking reliable alternatives and to manufacturers balancing expansion with financial discipline.

For Intel, the next important evidence will come from the relationship between construction progress and customer commitments. Arizona’s equipment installation and Ohio’s structural work show that the projects are advancing, while AI processor demand offers an immediate source of factory use. The longer-term question is whether external customers move from testing Intel’s processes to placing sustained production orders. That transition will help determine whether Intel’s expanding footprint becomes a meaningful second source for the industry or remains primarily an investment in its own product capacity.